题　目：Advanced Computational Topics for Integer Programming

嘉　宾：Yongzheng Dai, Ph.D Candidate, The Ohio State University

主持人：邓琪  副教授 金沙威尼斯欢乐娱人城

时　间：2024年11月28日（周四）14:00-15:30

地　点：金沙威尼斯欢乐娱人城 徐汇校区 安泰浩然楼308室

内容简介:

Integer Programming (IP) is a foundational NP-Hard problem that is widely applied but notoriously challenging to solve efficiently. Recently, accelerating IP solution processes is a popular topic in both academia and industry. This talk begins by addressing “Why IP?” — demonstrating the wide existence of integer programming models in real world and the potentiality to use IP to solve real-world problems. The discussion then shifts to leveraging parallel computing with multi-core CPUs to enhance IP solver performance, highlighting two recent parallel algorithms for improving general mixed-integer linear programming (MIP) solvers. Finally, we explore opportunities for further computational improvements in IP, including using random projection technique for mixed-integer (non)linear programming (MI(N)LP), developing feasibility-guaranteed neural networks for MINLPs, and some potential directions in global optimization.

演讲人简介：

Yongzheng Dai is a Ph.D Candidate in Operations Research at Integrated Systems Engineering, The Ohio State University. His research focuses on designing and implementing (parallel) algorithms for large-scale and mixed-integer (non)linear programming problems. Specifically, he is interested in 1) Designing (parallel) methods to improve solvers for MIPs and MINLPs; 2) Constructing efficient relaxations for complex global optimizations; 3) Developing decomposition or approximation methods for large-scale real-world problems.

欢迎广大师生参加!